Summary
Amodei said pacing would not halt model training or technical progress. Instead, companies would take more time to align and safeguard models while third-party evaluators verify their claims.
Why it matters
The clarification narrows the proposal from a general development freeze to a governance mechanism focused on verification and controlled release speed.
Related coverage
- Anthropic calls for pacing frontier AI and permanent embedded evaluators
- OpenAI commits to independent evaluators with employee-like access
- Sam Altman hints at an inter-lab AI safety pact